Skip to content

GitLab

Switch branch/tag
  1. 13 Dec, 2022 4 commits
    • dhansmair's avatar
      in the resize() function in image_transforms.py, the line 267: (#20728) · 30d8919a
      dhansmair authored 
      `image = to_channel_dimension_format(image, ChannelDimension.LAST)`
is redundant as this same conversion is also applied in to_pil_image().

This redundant call actually makes the training fail in rare cases.
The problem can be reproduced with the following code snippet:
```
from transformers.models.clip import CLIPFeatureExtractor
vision_processor = CLIPFeatureExtractor.from_pretrained('openai/clip-vit-large-patch14')
images = [
    torch.rand(size=(3, 2, 10), dtype=torch.float),
    torch.rand(size=(3, 10, 1), dtype=torch.float),
    torch.rand(size=(3, 1, 10), dtype=torch.float)
]
for image in images:
    processed_image = vision_processor(images=image, return_tensors="pt")['pixel_values']
    print(processed_image.shape)
    assert processed_image.shape == torch.Size([1, 3, 224, 224])
```

The last image has a height of 1 pixel.
The second call to to_channel_dimesion_format() will transpose the image, and the height
dimension is wrongly treated as the channels dimension afterwards.
Because of this, the following normalize() step will result in an
exception.
      30d8919a
    • Matt's avatar
      Fix AdamWeightDecay for TF 2.11 (#20735) · 4f1788b3
      Matt authored 
      * Fix AdamWeightDecay for TF

* Fix AdamWeightDecay for TF

* make fixup
      4f1788b3
    • Yih-Dar's avatar
      Change a logic in pipeline test regarding TF (#20710) · a12c5cbc
      Yih-Dar authored 
      

* Fix the pipeline test regarding TF

* Fix the pipeline test regarding TF

* update comment
Co-authored-by: default avatarydshieh <ydshieh@users.noreply.github.com>
      a12c5cbc
    • Younes Belkada's avatar
      Add `keep_in_fp32_modules` support (#20683) · 1af4bee8
      Younes Belkada authored 
      

* add `keep_in_fp32_modules` support

* pass it as class attribute

* few modifs

- make tests `slow`
- fix logic

* better logic

* fix failing test

* `bfloat16` support

* Update src/transformers/modeling_utils.py
Co-authored-by: default avatarSylvain Gugger <35901082+sgugger@users.noreply.github.com>

* fix

* simplify tests

* simplify tests

* fix test

* modify message

* more checks

* fix failing tests

* add more conditions

- add `is_accelerate_available`
- fixes pipleine tests that failed

* add suggestions

* Update src/transformers/modeling_utils.py
Co-authored-by: default avatarSylvain Gugger <35901082+sgugger@users.noreply.github.com>

* fix failing `bnb` test

* add last safety checker
Co-authored-by: default avatarSylvain Gugger <35901082+sgugger@users.noreply.github.com>
      1af4bee8
  3. 12 Dec, 2022 18 commits
  5. 09 Dec, 2022 7 commits
  7. 08 Dec, 2022 11 commits